History of Seaside Sparrow subspecies in Texas

It would be difficult to improve on the taxonomic and distributional history of the species presented by Robbins (1983), so we give only a brief summary relevant to Texas here. Allen (1888) first described sennetti from Corpus Christi, Texas. Ridgway's (1896) statement that macgillivraii ranged to the coast of Texas “during migration” was based on Audubon's (1838) extension of the range to include Texas, although Audubon gave no evidence that he examined any Texas specimens (macgillivraii was later understood to occur only on the southern Atlantic coast). Chapman described fisheri from Grand Isle, Louisiana, and characterized its range as “probably to northeastern Texas, southward in winter to Corpus Christi, Texas” (Chapman 1899:11). Following AOU (1910) in extending the northward range of sennetti to Galveston, Griscom and Nichols (1920:20) described the range of sennetti as “Coast of Texas from Galveston south to at least Brownsville.” Of fisheri, they wrote, “Breeds from Grand Isle, La., westward and south at least to High Isl., and possibly to Tivoli, Texas” (Griscom and Nichols 1920:24). Thus, by the 1920s we have fisheri occupying most of the Texas coast, from Louisiana to somewhere near Corpus Christi, and sennetti occupying an area around Corpus Christi, possibly north to Galveston, and perhaps making southward movements to the Rio Grande in winter.

Griscom later expanded on our understanding of Seaside Sparrows in Texas in 2 papers. He first argued that all Seaside Sparrows along the Gulf Coast are “positively proved to be strictly resident on their breeding grounds” (Griscom 1944:327). In the same paper, Griscom also noted that sennetti was apparently “extirpated at Brownsville. . .” (p. 323). In his subsequent work, Griscom concluded that perhaps earlier collectors in southern Texas had not been meticulous in their labeling of specimens and that the collecting locality of “Brownsville” might be doubted: “Thirty years of field work and search have failed to discover a breeding colony of Seaside Sparrows south of Nueces Bay” (Griscom 1948:107). Griscom did not comment again on his earlier statements about extirpation from Brownsville, but did add that any record from “near the Mexican border. . .must have been a straggler on migration” (Griscom 1948:108).

The most recent AOU checklist to formally treat subspecies described the range of sennetti (presumably on the basis of Griscom's works) as “Resident in coastal marshes (Nueces and Copano bays [Texas]). In winter south to the mouth of the Rio Grande” (AOU 1957:596). Oberholser (1974) noted 2 specimens of sennetti from Cameron County, writing that they were “Taken outside breeding range...” (Oberholser 1974:912). Lastly, in the interest of completeness, Oberholser (1974) reported a specimen of “howelli” from Jefferson County, Texas, but Griscom—who originally described howelli from Alabama (Griscom and Nichols 1920)—had later invalidated howelli after examining additional specimens (Griscom 1948). We note that, so far as we can tell, none of the early workers provided any evidence of regular seasonal migration beyond an assumption that records during nonbreeding seasons were, or could be, migrants.

Year-round occurrence and nesting of sennetti in Cameron County, Texas

Cameron County is the southernmost coastal county in Texas; it is bordered to the south by the Rio Grande and Mexico. In 1999 MHC captured a juvenile Seaside Sparrow during routine mist-netting at Laguna Atascosa National Wildlife Refuge (NWR) in Cameron County. In the same year, Phillips and Einem (2003) reported circumstantial evidence of breeding Seaside Sparrows near the Rio Grande in Cameron County. Nesting was subsequently documented in 2012 at Laguna Atascosa NWR (Ferrato et al. 2014). Subsequent across-season and across-year band encounters (N = 20; MHC, unpubl. data) of Seaside Sparrows banded at Laguna Atascosa NWR, seasonal density estimates (Ferrato et al. 2017), and many observations year round (MHC, unpubl. data) demonstrate that Seaside Sparrows are present year round, and breed at Laguna Atascosa NWR.

Habitat of Texas Seaside Sparrows in Cameron County

The habitat occupied by Texas Seaside Sparrows in Cameron County consists mainly of Borrichia frutescens (sea ox-eye) and Batis maritima (salt-wort), and it is notably different from the salt marsh (dominated by Spartina, Juncus, Distichlis) used by Seaside Sparrows elsewhere along the Gulf Coast (Ferrato et al. 2017). We conducted a first attempt to estimate the extent and distribution of potential Seaside Sparrow habitat in the 5 southernmost coastal Texas counties (Nueces, Kleberg, Kenedy, Willacy, and Cameron) using soil and vegetation data in ArcMap 10.3.1 (ESRI, Redlands, California, USA). We first plotted all points at least 100 m apart (N = 20; MHC, SW) in Cameron County known to have Seaside Sparrows. We gathered soil data from the United States Department of Agriculture (USDA) Web Soil Survey ( https://websoilsurvey.sc.egov.usda.gov/) and vegetation community classifications from Texas Parks and Wildlife Ecological Mapping Systems ( https://tpwd.texas.gov/landwater/land/programs/landscape-ecology/ems/). Soil and vegetation data were clipped to exclude areas > 10 km from the coast as we have no evidence that Seaside Sparrows occur that far inland in the region. We then intersected soil and vegetation layers to produce a map that highlighted areas of potential habitat (i.e., areas that matched both the soil types and habitat classifications of the bird points in Cameron County) in all 5 counties.

Population genetics and genetic effective population size of Seaside Sparrows in Texas

We revisited the population genetic analyses of Woltmann et al. (2014) to incorporate older DNA samples collected by CH and M. Whitbeck from Cameron County (2005–2006, Laguna Atascosa NWR, N = 24), Aransas County (2004–2005, Aransas NWR, N = 3), and Chambers County (Anahuac NWR, 2004–2005, N = 49). We focus here on analyses using the Bayesian clustering program STRUCTURE (Pritchard et al. 2000, Hubisz et al. 2009), and generate new estimates of genetic effective population size (N_e) using temporal approaches not possible earlier. Lab procedures (DNA extraction, PCR, and genotyping at microsatellite loci) were identical to those in Woltmann et al. (2014), and several samples from that study were run alongside the Hill/Whitbeck samples to ensure genotyping consistency. One locus from the previous study (Asµ15) did not amplify consistently in the older samples from 2004–2006, thus the analyses here are based on 13 microsatellite loci.

We identified 6 potential biologically meaningful groups a priori, distinguishing “old” samples (2004–2006) from “new” ones (2012), and separating Cameron County samples into “breeding” season (May–Aug) and “non-breeding” (Oct–Feb); samples from other localities were all “non-breeding.” In STRUCTURE, we used an admixture model with correlated allele frequencies, using our group designations as priors in the loc prior option, and otherwise default settings. STRUCTURE parameters were: burn-in = 100,000, number of replications after burn-in = 1.5 × 10⁶, and 15 iterations at each K (number of clusters) from 1 to 6. STRUCTURE output was first processed by STRUCTURE HARVESTER (Earl and vonHoldt 2012), and we used CLUMPP (Jakobsson and Rosenberg 2007) and DISTRUCT (Rosenberg 2004) to produce bar plots of cluster membership.

We estimated N_e for various populations using 3 approaches: (1) sibship frequency (SF), (2) linkage disequilibrium (LD), and (3) a temporal approach. The SF approach was implemented in the program Colony (Wang 2016), using a model based on polygamy in both males and females and otherwise default settings. The LD (lowest allele frequency = 0.05, parametric 95% confidence intervals) and temporal (Jorde/Ryman method, parametric 95% confidence intervals) approaches were both implemented in NeEstimator (Do et al. 2014).

Habitat of Texas Seaside Sparrows in southern Texas

In Cameron County, all bird locations occurred on Barrada Clay. Most locations were classified as Salt and Brackish High Tidal Marsh, but Sea Ox-eye Daisy Flats and Salty Prairie were also represented. In Cameron and Willacy our approach identified ∼11,000 ha of potential habitat, but identified <40 ha in Kenedy, Kleberg, and Nueces combined (Fig. 1). We contrast the distribution of this potential habitat with the distribution of typical Seaside Sparrow habitat (displayed in Fig. 1 as Texas Coast Salt and Brackish Tidal Marsh, including both high and low tidal marsh), which is described as being dominated by Spartina alterniflora (smooth cordgrass) and Juncus roemerianus (black needlerush), and not infrequently containing Avicennia germinans (black mangrove;  https://tpwd.texas.gov/landwater/land/programs/landscape-ecology/).

Figure 1

Potential distribution of Seaside Sparrow habitat along the Texas coast. Population genetic sampling localities are Cameron, Aransas, and Upper Coast.

Brief reconnaissance surveys in March 2017 (SW, AX, MR, and MHC) in Cameron and Willacy counties indicate that this first estimate of available habitat is too generous. Although we did find previously unreported Texas Seaside Sparrows in Willacy County, most of the areas we were able to visit did not appear to support Seaside Sparrows, and refinement of the habitat models is clearly needed. Barrada Clay is largely absent from the region between Willacy and Nueces counties. Barrada Clay is present in Nueces and Aransas counties (around Nueces, Copano, and Aransas bays), but the vegetation community that supports Seaside Sparrows there is classified as Salt and Brackish Low Tidal Marsh.

Population genetics and genetic effective population size

Based on log likelihood scores (Fig. 2; and also the approach of Evanno et al. 2005, not shown), the most conservative interpretation is that there are 2 clusters (K = 2), although K = 3 has very similar support. Cameron County samples form one cluster, separate from all other samples (K = 2), and if K = 3, Aransas samples are additionally separated from Upper Coast samples (Fig. 3). These analyses suggest a closer affinity to fisheri in samples from Aransas than was found in previous analyses (Woltmann et al. 2014), although a signal of admixture with Cameron County samples is still present. The K = 3 solution is on one hand more challenging to interpret because there is no obvious barrier between the marshes around Aransas and Corpus Christi and marshes to the north. Oberholser (1974) commented on extensive marsh destruction along the Texas coast, but it is not clear if this has created or contributed to a dispersal barrier for Seaside Sparrows. On the other hand, in the absence of the Cameron County samples analyzed here and by Woltmann et al. (2014), the underlying population genetic signal would more or less align with the commonly accepted ranges of sennetti and fisheri.

Figure 2

Mean likelihoods for the number of clusters (K) as estimated by STRUCTURE and plotted by STRUCTURE HARVESTER. We interpret K = 2 as the most conservative solution, but note that K = 3 also has a reasonable biological interpretation.

Figure 3

STRUCTURE bar plots considering either 2 or 3 clusters (K). “Cam.” = Cameron County, “Aran.” = Aransas County, “Upper” includes Brazoria County (2012) and Chambers County (2005). All samples are from the nonbreeding season (Oct–Feb) except “Cam. 2005” (May–Aug).

Estimates of N_e using the LD approach were mostly not useful, with point estimates or upper confidence limits not infrequently infinite (Table 1); this is generally interpreted as a weak linkage disequilibrium signal (Do et al. 2014). Estimates of N_e using the SF and temporal approaches were qualitatively consistent: Cameron County estimates were approximately half of the Upper Coast estimates within each method.

Table 1

Estimates of genetic effective population size (N_e, with 95% confidence intervals) of populations of Seaside Sparrows in Texas. Estimates were generated in the programs Colony (Sibship Frequency approach) and NeEstimator (Linkage Disequilibrium and Temporal approaches).

Future work

Most of what we know about Texas Seaside Sparrows at the southern end of their range comes from a single site at Laguna Atascosa NWR. Because of our poor understanding of the distribution of Seaside Sparrows in the region, we lack estimates of overall population size and have no information about population trends. Threats to this population include sea level rise and encroachment of black mangrove, both of which will eliminate habitat for these birds. At least one formerly occupied area on the refuge is now 100% mangrove, and the sparrows no longer occur there (MHC, pers. obs.). This creates an unfortunate potential conflict between 2 native species of conservation concern (i.e., black mangrove and Texas Seaside Sparrow), although it should be noted that the mangrove seems to be expanding its range due to climate change (Montagna et al. 2011). Along with a more formalized and sophisticated plan to map the distribution of the sparrow and to monitor population size over time, we suggest that both sparrow habitat expansion methods and mangrove control measures in areas currently occupied by the sparrow should be explored.

The habitat distribution analysis presented here is an oversimplification, and more work is needed to better understand habitat use and distribution along the Texas coast. For example, although much of the marsh habitat around Corpus Christi and Aransas is classified as Salt and Tidal Low Brackish Marsh, the plant species composition (e.g., Borrichia, Batis) is often similar to what is found farther south, albeit in a wetter context. Additional work is needed to understand the role of soil type, context, and distribution, especially with regard to Barrada Clay and its influence on vegetation composition. The region between Corpus Christi and Houston encompasses an aridity and salinity gradient that includes a transition zone from marshes dominated by woody or succulent plant species to more typical graminoid-dominated marshes (Osland et al. 2014, Gabler et al. 2017). Whether the genetic signal of introgression between northern and southern populations of Seaside Sparrows along the Texas coast tracks this habitat gradient remains unknown for now.

Griscom (1948) commented on the clinal nature of specimens from Corpus Christi to the northeast, and finer-scale (both geographic and temporal) genetic sampling throughout the region is needed to better characterize the intergrade zone. We are not ready to dismiss entirely earlier workers' discussions of migration of either fisheri or sennetti; inferences of seasonal movements in Mississippi in fall (M. Woodrey, unpubl. data) and movements related to hurricanes (Stouffer et al. 2013) both suggest that landscape-scale movements are possible, if not probable. A better understanding of seasonal movements by either group could certainly impact (and inform) population genetic inferences. Higher-resolution genetic data (e.g., single nucleotide polymorphisms) would be very useful to compare the behavior of neutral markers to those under selection in this region.

A new look at old specimens is needed to ask how well the current population genetic inferences match those made from specimens. Based on observations made during field sampling, birds sampled in Aransas County are certainly paler overall and have a more olive dorsal cast than typical fisheri from elsewhere along the Gulf Coast, but the birds in Cameron County seem paler still (SW, pers. obs.). Quantitative plumage data are needed. Nearly all earlier workers have commented on the extent of within-population variation in Seaside Sparrows (with which we wholeheartedly agree), but we wonder if some of the confusion described by Griscom in the Corpus Christi area was caused by his assumption that either (1) birds labeled “Brownsville” could not have been collected farther south than Nueces Bay, or (2) birds collected from south of Nueces Bay could not possibly differ from the birds he examined around Corpus Christi.